Snow is a vital environmental parameter that holds significance across various disciplines, such as hydrology, meteorology, and natural disaster management. With the increasing accessibility of snow products derived from Synthetic Aperture Radar (SAR) and optical data, like Sentinel-1 wet snow and Sentinel-2 total snow, users have benefited from improved snow mapping and monitoring. However, snow mapping in the mountainous areas remains challenging due to the difficulty of obtaining reliable ground truth data on steep mountain terrain. In this study, we introduce a deep semantic segmentation framework, SACUNet, specifically designed for wet snow detection from SAR image time series in mountainous environments. To address the lack of ground truth, we constructed a high-confidence training and validation database through a rigorous decision-fusion process combining multi-temporal Sentinel-1 wet snow detections with Sentinel-2 total snow maps. We also propose two complementary metrics, the Conditional Agreement Rate (CAR) and the Wet Snow Intersection over Union (WSIoU), to quantify the robustness and consistency of the fusion procedure, therefore ensuring the reliability of training labels in the absence of in-situ data. SACUNet integrates advanced techniques like: (i) Depthwise Separable Convolution, which captures cross-channel dependencies and adapts feature representations, and (ii) Atrous Separable Convolution, which further refines and consolidates the learned features, into the U-Net architecture. The proposed framework has been successfully employed to monitor wet snow in the Mont-Blanc massif, using a time series of 69 Sentinel-1 images acquired from 05 July 2020, to 29 August 2021. SACUNet demonstrates remarkable accuracy in wet snow detection, with an Overall Accuracy of 97%, Precision of 94%, Recall of 97%, Intersection over Union at 92%, and an F1-Score reaching 96%. Validation against meteorological records from four alpine stations confirmed that SACUNet effectively tracks seasonal wet snow dynamics, suppresses false detections during cold periods, and captures realistic high-altitude melt events. Moreover, the model trained in Mont-Blanc generalized successfully to the Vanoise massif, demonstrating its transferability to other alpine regions. Beyond quantitative accuracy, SACUNet enables the spatio-temporal analysis of wet snow evolution, offering insights into its extent, frequency, and seasonal progression across elevation bands. These findings highlight the framework’s potential as an operational tool for large-scale wet snow monitoring in mountainous environments.

Snow holds a significant importance as a fundamental environmental factor in multiple domains. Obtaining accurate ground truth data for snow mapping in mountainous areas presents a significant challenge. To address this issue, this paper presents a deep semantic learning framework for the segmentation of Sentinel-1 images for wet snow detection in mountainous areas. Firstly, we propose to create a deep leaning database based on snow products derived from Sentinel-1 and Sentinel-2 data. Afterward, we introduce a deep convolutional neural network called ReXcepUnet, which combines the U-Net architecture and the powerful Xception backbone. Finally, the proposed framework has been successfully applied to monitor wet snow in the Mont Blanc massif, yielding high accuracy results. The ReXcepUnet model demonstrates a good performance in wet snow detection, particularly in high-relief regions like the Mont Blanc massif.

Recent, an advanced Persistent Scatterers and Distributed Scatterers (PSDS) InSAR algorithm has been implemented as an open-source TomoSAR package (https://github.com/DinhHoTongMinh/TomoSAR). This effort aims to contribute the spatial distribution of subsidence in Ho Chi Minh City (HCMC), the most crowded city and economic hub of Viet Nam, in its horizontal and vertical components by using TomoSAR. With Sentinel-1 data, taking into account the presence of east-west horizontal motion, our findings indicate that the accuracy of the decomposed vertical velocity can be improved by up to 3 mm/year for Sentinel-1 data. The obtained results revealed that subsidence is most pronounced in the areas along the Sai Gon River, in the northwest-southeast axis, and in the southwest of the city, with a maximum value of 80 mm/yr, which is in accordance with the findings of the literature. The amplitude of east-west horizontal velocities is relatively small and large-scale eastward movement can be observed in the west of the city at a rate of 3–5 mm/year. This confirmed that the displacement in Ho Chi Minh City area is mainly vertical downward.

SAR tomography, a technique employing multiple acquisitions over the same areas to form a three-dimensional image, has been demonstrated to improve SAR's capability in many applications. Our study shows the potential value of SAR tomography acquisitions to improve forest classification. By using P-band tomographic SAR data from the German Aerospace Center F -SAR sensor during the AfriSAR campaign in February 2016, the vertical profiles of five different forest types at a tropical forest site in Mondah, Gabon (South Africa) were analyzed and exploited for the classification task. We demonstrated that the high sensitivity of SAR tomography to forest vertical structure enables the improvement of classification performance by up to 33 %. Interestingly, by using the standard Random Forest technique, we found that the ground (i.e., at 5–10 m) and volume layers (i.e., 20–40 m) play an important role in identifying the forest type. Together, these results suggested the promise of the TomoSAR technique for mapping forest types with high accuracy in tropical areas and could provide strong support for the next Earth Explorer BIOMASS spaceborne mission which will collect P-band tomographic SAR data.

This paper addresses the monitoring of volcanic eruption using a coherence change detection matrix constructed from a multitemporal InSAR image time series. The Piton de la Fournaise volcano (French island, La Reunion), one of the most active volcanoes worldwide, was selected as a case study. Changes on the ground related to eight volcanic eruptions were analyzed through a time series including 49 descending stripmap Sentinel-1 SAR images acquired from January 10, 2018 to August 21, 2019. The experimental results have shown the relevancy of the proposed framework.

This paper presents a multiscale framework for change analysis using a big dataset of multitemporal Synthetic Aperture Radar (SAR) images. In patch scale, spatio-temporal change information can be extracted by the Patch-based Change Detection Matrix (P-CDM). Then regions and acquisitions of interest that show where and when changes occurred are determined from obtained P-CDMs. In pixel scale, for each such region, changes are detected in details between selected acquisitions using statistical similarity measure. This approach was then applied for analyzing changes along central coast of Vietnam (from Hue to Quang Ngai) by using a time series including 19 Sentinel-1 images. The experimental results have shown the relevancy of the framework in detecting abrupt and seasonal changes, and its effectiveness in processing a long time series with large size images.

International audience

This paper first provides statistical properties of wavelet operators when the observation model can be seen as the product of a deterministic piecewise regular function (signal) and a stationary random field (noise). This multiplicative observation model is analyzed in two standard frameworks by considering either: 1) a direct wavelet transform of the model; or 2) a log-transform of the model prior to wavelet decomposition. The paper shows that, in Framework 1, wavelet coefficients of the time series are affected by intricate correlation structures which blur signal singularities. Framework 2 is shown to be associated with a multiplicative (or geometric) wavelet transform, and the multiplicative interactions between wavelets and the model highlight both sparsity of signal changes near singularities (dominant coefficients) and decorrelation of speckle wavelet coefficients. This paper then derives that, for time series of synthetic aperture radar data, geometric wavelets represent a more intuitive and relevant framework for the analysis of smooth earth fields observed in the presence of speckle. From this analysis, this paper proposes a fast-and-concise geometric-wavelet-based method for joint change detection and regularization of synthetic aperture radar image time series. In this method, geometric wavelet details are first computed with respect to the temporal axis in order to derive generalized-ratio change images from the time series. The changes are then enhanced, and speckle is attenuated by using spatial block sigmoid shrinkage. Finally, a regularized time series is reconstructed from the sigmoid shrunken change images. Some applications highlight relevancy of the method for the analysis of SENTINEL-1A and TerraSAR-X image time series over Chamonix Mont Blanc.

This paper presents a method for analyzing SAR image time series and provides initial change detection results on a time series of 11 descending Interferometric Wide Swath (IW) Level-1 Single Look Complex (SLC) Sentinel-1 SAR images over Chamonix-Mont-Blanc, France. This method is based on the Change Detection Matrix (CDM) which identifies the presence of changes in the time series. It provides a useful information to gather homogeneous samples for spatio-temporal speckle filtering and to obtain a map of change dynamics in order to reveal the temporal evolution.

This paper proposes an approach for temporal adaptive filtering of Polarimetric Synthetic Aperture Radar (PolSAR) image time series by integrating a change detection technique. The filtering strategy is based on the detection of changed and unchanged areas derived by applying an appropriate similarity test. A time series including 7 descending fine-quad polarization RADARSAT2 images acquired from January 29, 2009 to Jun 22, 2009 over Chamonix-MontBlanc test-site which includes different kinds of change is used to validate the proposed method.

This letter presents an adaptive filtering approach of synthetic aperture radar (SAR) image times series based on the analysis of the temporal evolution. First, change detection matrices (CDMs) containing information on changed and unchanged pixels are constructed for each spatial position over the time series by implementing coefficient of variation (CV) cross tests. Afterward, the CDM provides for each pixel in each image an adaptive spatiotemporal neighborhood, which is used to derive the filtered value. The proposed approach is illustrated on a time series of 25 ascending TerraSAR-X images acquired from November 6, 2009 to September 25, 2011 over the Chamonix-Mont-Blanc test-site, which includes different kinds of change, such as parking occupation, glacier surface evolution, etc.

The paper addresses multi-channel complex image filtering. It provides regularization cost functions associated to non-conventional vector and matrix iv norms for promoting geometry properties. The approach is shown to be efficient for filtering PolSAR images.